How Amazon sellers can separate AI gold from garbage

This This is the Billiondoll Sellers podcast. Your go-to source for cutting edge strategies and success stories from the world of Amazon and e-commerce. Buckle up and get ready to take your Amazon business to new heights. Don't forget to subscribe to the Billiondoll Sellers Newsletter. Welcome your host. >> Welcome your host, Kevin King. >> Hey everyone, and welcome to the Billiondoll Sellers podcast. I'm your host, Kevin King, and today is Monday, June 29th, 2026. We got a packed show today, so let's jump right in. So, today we're going to talk about how to fix your listing categorization on Amazon because BDSS Dream 100 member Vanessa Hung breaks down how a lot of sellers are quietly losing traffic and they don't even know why. Um, then we've got some interesting stats on the global e-commerce landscape. Amazon's new AI disclosure rules for your A+ content images. Uh, Prime week numbers are in and a really cool breakdown of this AI tool called Higsfield that promises to build your entire Amazon brand from one prompt, plus the usual hot picks and importing shots. All right, here's your stump Bezos question for today. So, by coding has completely changed the app landscape. 26% of all apps launched in 2025 are already gone. The question is, what percentage of apps launched back in 2023 are gone now? Think about that and I'll give you the answer at the end of the show. But first, a quick shout out. My buddy Jr. Galli's running his Cloud Code Challenge again. Uh, it's called Build Your First AI Employee in 5 days. It actually starts today, June 29th. And if you missed the start, don't worry. Replays are available. He walks you through building AI automations with zero code, connecting stuff like Gmail, Google Calendar, Slack, Asa, all that kind of thing. If you've been wanting to get into AI automation, but you're not a coder, this is a great place to start. There's a link in the show notes to check it out. All right, let's get into it. So, this one's a double hack today, and it comes from Vanessa Hung. She's the founder and CEO of Online Seller Solutions and BDSS Dream 100 member. And what she's saying is, you know, you got to get your listing on the right shelf first, and then you got to figure out how Amazon decides whether it let you keep it there. So, let's start with part one, which is fixing your categorization before it quietly kills your traffic. And in the AI era, your data quality is basically your product quality. You can have a great title, killer bullets, beautiful A+ content, but none of it matters if Amazon has you filed on the wrong shelf. Whether it's Alexa for shopping or A9 or whatever's rooting the customer, if Amazon can't categorize your product, they can't show it to anybody. So, there's three attributes that decide where your listing actually lives. You got the browse node, which is your overall category. You've got the product type or prototype, which defines which attributes your listing needs. And then the item type keyword, which is the one that matters most right now. And Vanessa has this great analogy. She says, "Think of the browse node as the highway. The prototype is the avenue, and your product is the little house on that avenue. And Alexa is the Uber driver. If the driver doesn't know what street you're on, you never get the ride. Doesn't matter how nice the house is." I really love that analogy. It just makes it click. Yeah, here's why this is urgent. This past February 26, Amazon restructured the browse tree guide through a flat file change. It quietly reshuffled prototypes and item type keywords and uh a lot of sellers got caught off guard. A lot of listings fell out of line and sellers have no clue. They're just watching their traffic slide and they don't know why. She gave an example where products showed air mattresses in the front end path, but the backend item type keyword said inflatable beds closed not fatal but not aligned. and her rule of thumb is basically the last word of your category path should match your item type keyword. When it doesn't, you leak traffic. Now, here's the gotcha that most sellers miss. Amazon's policy now says your title, bullets, and description are key attributes used for product classification. You used to pick your category. Now, Amazon reads your copy and picks for you. So, the fix isn't just a backend flip. You might have to rewrite your copy to push the algorithm to the right category. So, what do you actually do about it? Here's the steps. First, check the back end directly. There's a URL on Solar Central. Uh, it's in the show notes that shows what Amazon really has stored for your ASM, and you can swap the domain.com to mx.ca. EU to check every market you sell in. Then pull your category listing report from seller central. Run it through flatfile transfer.com, which is a free tool that Vanessa built. It flags every discrepancy and tells you the exact item type keyword your flat file should carry. Download the category specific blank template from Amazon's catalog tools for the right item type. Grab one for every keyword you need. And she uses a cloud skill to auto transfer the data between the report and the template so you're not retyping everything by hand. Upload, map, adjust, push. All right. Now, part two, and this is where it gets really interesting. Vanessa doesn't stop at fixing it. She says you got to make sure your edits actually win. Because here's the catch. You can fix all of this and still get overwritten. And it only makes sense once you know who Amazon lets changes. So the Amazon catalog runs on what's called a contribution hierarchy. Every contributor gets a score from 0 to 100 and the highest score wins every time. No exceptions. And here's the breakdown. Sellers without brand registry get 30 points and the catalog team sits above them. Sellers with brand registry get 52 points with the brand registry team above them. The Amazon data augment sits at 52.1, just barely above brand registry sellers. And vendors get 60 points with the retail team above them. And Amazon's internal teams sit at 100 points at the top. So every attribute on your listing has one authoritative owner. Title, brand name, bullets, images, each one. When two contributors submit different values for the same field, Amazon doesn't negotiate. It compares scores, keeps the higher one, rejects the other one automatically. Think of your listing as a leaderboard. If you're not the top score on a field, you're locked out of it. Somebody else owns it. And that's why the brand registry jump matters so much. It lifts you from 30 to 52. And that client puts your edits above the open catalog. The points you earn are the points that protect your business. Um, so the takeaway here is audit your backend item type keyword against your frontend category path. If they don't match, your copy is routing you to the wrong shelf and uh nor PPC fixes a listing the algorithm can't place. Then check your contribution score. Fixing your listing only counts if you outrank everyone else trying to change it. Now, let's look at some interesting stats real quick. So, there's a chart going around from Market Maze showing the largest global e-commerce platforms by GMV in 2025 and uh the projected growth for 2026. Amazon still number one at $845 billion, projected to grow about 5.6%. But here's what's interesting. The established marketplaces still lends on scale, but the discovery platforms own the growth column. Tik Tok shop Tiny at 66 billion but is projected to grow almost 60% in 2026. That's insane. TMU is at 73 billion growing over 13%. AliExpress at 76 billion growing about 10.5%. And Dian, the Chinese version of Tik Tok is at $582 billion growing almost 13%. Meanwhile, Walmart's at $197 billion growing almost 9% which is, you know, solid for their size. And then eBay is basically flat at $74 billion, up like 1.6%. So the traditional marketpl, but the newer discovery driven platforms are where all the growth energy is going. Now, here's something you need to be aware of. Amazon just added a new disclosure step for your A+ content. When you upload images to A+ content or brand stories, you now have to tell Amazon two things. Is the image AI generated? And does it show a photorealistic AI person? So, there's two check boxes now when you upload. One for AI generated and one for AI generated people and Amazon captures all of it as metadata behind the scenes. But here's the thing. Shoppers see nothing right now. There's no AI generated label on the listing yet. And this only hits A+ content brand stories. Your main product images are untouched for now. So, does this actually change anything today? Uh, not really. Nobody's getting suppressed over it right now. But pay attention to the direction here. Amazon's building a paper trail on AI content. Once they have the data, they have options. Ranking signals, compliance rules, labels shoppers can see. None of that exists today, but the plumbing is going in. So, if you're leaning on AI for your creative, keep track of what you flagged and where. The disclosure is easy right now, but the rules that come later might not be. All right, the prime wing numbers are rolling in, and there's some really interesting data here. So, the top items by unit sold were premier protein shakes, liquid IV packets, temptations cat treats, Dawn power wash spray, and hefty ultra strong trash bags. And uh look at that list for a second. That's not flashy electronics. That's everyday consumables. That tells you a lot about what people are actually buying during Prime. Average price for item was $2323. And 26% of items were priced under $5. Another 16% under $10. So, we're talking about mostly low price point purchases. And what's really telling is why people bought. 45% said they've been waiting to buy something on sale. 36% said these were items they normally buy on Amazon anyway. 32% picked up general Prime Day deals. And 26% were just stocking up on sale items. So, about half of shoppers are using Prime as their chance to finally pull the trigger on stuff they've had their eye on. The top categories were apparel and shoes at 30%, household essentials at 28%, health and wellness at 27%, beauty at 26%, and homegoods at 23%. So, if you're selling in any of those categories, you know, that's where the money went this prime week. All right. Now, this next one is uh I think the most important story in today's show, and it's about separating the gold from the garbage when it comes to AI tools for Amazon sellers. So, there's a tool called Hicksfield Supercomputer making the rounds and one of its three demo workflows is aimed straight at you. The pitch is one chat, one prompt, a finished Amazon brand. So, in the demo, the operator types something like this. Suggest three unbranded Amazon products worth selling. Include price, margin, and competition. And the agent just browses bestseller list in real time, runs FBA fee math, reads reviews to find gaps, and hands back free picks. It lands on dog socks. $19 retail, a buck to make, 71% margin, 68,000 monthly searches. Then it names a brand, builds a product sheet, generates listing images, studies winning ads on meta, Tik Tok, and YouTube shorts, and writes 15 ad concepts with video to match all in one window. I mean, slid demo. Now, let's separate the gold from the garbage because this is the whole story of AI for Amazon right now. Watch the product research step again. The agent finds dog socks. $19 sale price, $1 cost, 68,000 searches, none with real branding, none with video ads. A virgin market hiding in plain sight. Think about that for one second. If an AI can surface that opportunity in 12 seconds from a generic prompt, so can the 50,000 other people who watched the same video and ran the same prompt this week, right? That low competition it's bragging about, uh, that goes away the second the tool exists. You're not early. You're the demo. And this is the AI hype trap in a nutshell. Well, the flashier the part of the workflow, the less edge it actually gives you. If this flashy means easy, and easy means everyone has it. Product research, niche picks, wearing product finders. That's the casino floor. The house already knows the dog's gold rush is coming. Now, the unsexy part that nobody clips for the highlight reel, asset production, listing image sets, infographics, comparison cards, lifestyle shots, 15 advancements in a sitting. This is the work that actually eats your weeks and drains your agency retainer. And it's the work AI generally compresses from days to minutes. And one real edge here is it actually watches competitor video ads with the audio instead of reading a text summary of them like most tools for modeling what's working in your niche. That's actually worth something. And the pattern holds across every AI tool you'll evaluate this year. The boring middle of the workflow is where the money is. The magic button book ends. That's where the hype lives. Now, two things it won't save you from. The output still has AI tells. Hands clipping through products, items morphing between cuts. Nothing ships unreed and none of is Amazon compliant out of the box. Main image on white. No badges or text on a hero claims you can actually back up. The agent doesn't know your category rules, so that's still on you. And the suspension doesn't care that a robot made the asset. Under the hood, it just routes between Claude GPT and Gemini per task. The model isn't the product, the rapper is. So bottom line, it's a lousy oracle but a useful factory. Use AI to build faster, but don't use it to think for you. All right, before we wrap up, a few more hot picks for you. Amazon's Prime Day spending has already passed Adobe's estimates, which uh you know tracks with what we talked about last episode. The US Postal Service is changing how it measures dimensional weight. So if you ship through USPS, keep an eye on that. While container prices have soared to a 22-month high and China to US shipping is now over $5,000 a container. That's going to hit a lot of people's margins. Tik Tok shops commission cuts are changing influencer economics. 36% of deliveries from Walmart stores now arrive in 3 hours or less. How cool is that? And creators are claiming new Google search profiles to help boost their visibility. Links to all those are in the show notes. And here's your parting shot for today. This one's from President Ronald Reagan. Entrepreneurs and their small enterprises are responsible for almost all the economic growth in the United States. I love that one, you know, because it's easy to forget sometimes when you're in the weeds fighting listings and PPC and all this stuff. But yeah, that's literally what you're doing. You're driving the economy. So, don't forget that. And finally, about that Stump Bezos question from the beginning. I asked what percentage of apps launched in 2023 are gone now given that 26% of 2025 apps are already toast. The answer is only 2%. Just 2%. So the newer the app, the more likely it disappears. That vibe coding wave is uh churning out a lot of stuff that just doesn't stick. All right, that's all for today, folks. I'll see you again on Thursday. This is Kevin King signing off from the Billiondollar Sellers podcast.